I did a little more research on the wheels thing. It seems I'm going to have to re-do the process yet again, but with armatures to get the wheels on the cars to turn. (The answer lies in the topic of animating wheels on cars - but not the railroad kind.) It might be a tad more trickier, but to get it done right that's what's needed to happen.

Seeing how the armature does it via transforms instead of needing scripts just goes to show how much a blender noob I am. :blink:

At least I'll admit it. :laugh:

You're welcome. :)

If anyone is curious, the results are like this:
http://www.youtube.com/watch?v=Ui-gSaI_du0

I still consider this halfway done though... Why?

I can't figure out getting wheels to roll properly while following the path. It doesn't matter if wheels have no visible texture or smooth geometry like in the linked example, but it would matter if you had wheels with texture or bits of stuff attached like drive wheels on a steam engine. Rigid body physics behaves a little wonky with constraints (at least I'm not having luck), and I'm nowhere up to speed on driver or Python scripted animation. I know part of it involves knowing the wheel diameter (easy enough via correct object dimension) to get circumference (using some circle math involving pi) and then path length in units (count 1 unit boxes arrayed to fit path to get close) to figure out number of rotations (path length divided by circumference), but the exact details is where I get lost. (Getting Blender to do all that for you. I sorta know what has to be done, unfortunately just not Python and/or drivers part.)

If only I it were possible to port the Sparrowhawke3D wheels modifier Carrara plugin to Blender. (It's not, but something equivalent would make life easy.)

Enough talk... I'll just show how to do it, even though I'm a bit clumsy with this software still.

http://youtu.be/bL00aXePHUA

In Cycles renderer, all your mats have to be set to use nodes. (Push that "use nodes" button next to the materials settings.) If you don't do that, then yeah, it's going to render white.

Also, nodes are pretty awesome for setting up procedural materials. You may not have all the texture sets of the internal renderer (still being developed), but shaders are rather decent once you figure them out. Also node layout is nice for figuring out what is going on as as far as how textures get generated and applied.

I managed to get a crude version of the idea working for a single rail car.

Front and rear bogies are individually set up with a follow path constraint that has follow curve turned on. Each bogie also has an empty axes set as a child offset at the same height above its center. The car is set as a child to the empty axis above the front bogie. The front empty axis is also set to point-at the rear one. Then set the offsets for the front and rear bogies under the follow path constraint. (Don't offset by just moving them, it throws them off the path.) It'll take some adjusting to make sure everything is aligned and pointing the right way. (Offset value in follow path constraint isn't same as global distances afaik.)

But if you do it all right, you'll have the wheel bogies follow the track almost perfectly, and the car will stay in line over the bogies. Animate by using settings on the curve itself under path animation.

You may have your work cut out for you in animating an entire train. But it's definitely possible to have it follow the rail in a fairly realistic manner.

I also tried some other way with bones and IK stuff but wasn't having much luck. The empties and regular ol' constraints seem to work with the least fuss. (Although there might be a better way. I'm still new at it.)

Might want to use constraints and have the bogies follow the path. May also have to do some math to offset the F-curve path position start and end for each. (Basically you get all your wheels following the path while being spaced as if they had the cars and engines attached. Make path slightly longer. Offset start and ends of each so they don't pile up when animated.)

If you can make a group of bogies follow the path at the correct intervals, then it should be a matter of constraints and/or rigging (with IK?) to attach cars to those bogies.

Sounds tricky and complicated (and it most likely is), but may be a more realistic looking way of doing it than making just the cars follow the track and not caring which ways the wheels point or trying to get them following the path after the fact.

Alternately bullet physics with rigid bodies may accomodate what you want to do as well. But then you run into issues of tolerances and weird simulation behavior with offsets and the occasional miscalculated collision. A physics simulation may be more intuitive, but the unexpected stuff might have you pulling more of your own hair out.

Haven't done it yet, but I can picture how it should work in my mind. I might have to try this outside of my all-too-short lunch break.

Darqmind... I think that's why there's a bleed setting. If your maps are too tight, some effects will blend in stuff from the edges. Bump the bleed up a bit when texturing to get more overlap and make sure there's room between non-adjacent UV charts so you can do that.

As for texturing in general in Blender, anyone got an anser for this?

http://www.youtube.com/watch?v=ENS6QQZIsgk

Only anwser I could think of is having a side project where I join the parts that share a map in order to texture paint the model. (Remembering not to save over the other project where the parts are all separate as they should be.) If only I could do what I wanted to, it seems like that shouldn't be necessary.

I came back to Blender after trying it a few times previously. It seems the 2.6x versions have finally done what's necessary to make the software a little more noob-friendly. (Hotkeys displayed on menus, majority of stuff now exists on menus so you don't have to know the hotkeys, tooltips, alternate software modes, etc. Almost all the kind of essential UI stuff that's there for hand-holding and to show newbies the way.)

In terms of coming from other software with easy UIs like Bryce and Cararra, that might mean something.

The Cycles renderer also seems a good step in the right direction too. Lighting environment in that renderer needs little to no fakery. Only problem is the occasional caustic-related "fireflies".

I think the only major thing it lacks is an in-depth preset library and browser. (Yes there are some plug-ins. But such feature should be integrated.) Add that and buff out a few of the rough spots in regards to workflow (I'd say they're nitpicks and gripes - of which I have a few, since the software is usable as it is), and there's no reason why Blender can't be a "pro" 3D graphics package.

Blender is almost where it needs to be. Once it is, the "bottom tier" commercial 3D software may have a real reason to worry. (Or they better start making good content and support deals. Blender is nibbling at them as it is.)

Cool. That seems to work well. Particle spawner and drop to floor? Or what?

Quote - Wouldn't it be fun to have the ability to add random image maps to a surface? 

I think that might be possible too. I'd try a numbered image sequence like you'd have for animated image maps. Then use the randomize option on the timeline slider for the frame number. Only catch is that I'd think each item would have to have a separate randomizer seed applied to its timeline, or they'd still end up all the same. That would take a bit of experimentation to be certain.

A gradient layer or two with masking and fidgeting around with blend mode settings might be one way to get multicolor lightning. Some of the other stuff looks a little complicated though.

Nice video. These things are always cool to watch. In addition to scenic skies with passing clouds and stars, cityscapes and cross-country drives are among my favorite timelapse subjects.

I know a little bit of technical on these too...

Don't mean to boast on Nikons, but if you've got one - many models have this thing called "Interval Mode" that makes this kind of thing relatively easy. With a tripod, it's something that may be neat to try out sometime. Of course it's still possible with other cameras, but they require extra kit in order to do the same thing. (Having it built in the camera makes more sense though, as almost all models have a clock of some sort built in anyways. Also this means less stuff to carry around and worry about if you do this kind of photography.)

Slow pans/dollys must take some planning or special kit though. Likely an astronomy tripod with tracking. Nudging a typical tripod manually over time works too, but is hard to get precise.

I've done one or two stabs at this kind of thing so far, and memory and batteries tend to go fast. Use external power if you can and avoid using the flash. Smaller image size and quality settings are good too, as photos from DSLRs are already much larger than HD video resolution to start out with and this gives the memory card a break. It's best to go full-manual and learn what settings to use, as auto-modes tend to jump around metering and white-balance in a way that looks funky. (Sometimes that look is interesting too, if you like a sort of grungy flickering feel to your time-lapse. Stop-motion in Tool or NIN music videos comes to mind.)

It's challenging to get right, but I find it fun to do. I really need to try doing some more sometime. The downside is that it typically involves "babysitting" the camera, particularly if you can't guarantee that it's location is secure from tampering or theft. Having a comfy portable chair and bringing your laptop or some books to read will make that easier.

Then you've got to figure out how to stitch still photos into video. Something like Sony Vegas or perhaps Adobe Premiere will work. This can also be done for free btw. (That's how I do it.) Script-fu batch interpreter in Gimp to crop/resize images to video size and save sequenced copies to a working folder, and Virtualdub to stitch the sequential frames from the working folder and converting to video while applying codec and compression. In regards to compression, Xvid seems to work best and with the least hassle while on a freebie budget. When on a PC, I definitely suggest using those as they work much better than MS Movie Maker for this kind of thing. I believe Macs also come with a good video editor already included, but iirc it costs extra to unlock some features.

Hmmm...  I was thinking of some other things too. Make sure the scanner is clean of any dust. Ditto with the drawings. Under image adjustments (or adjustment layers), you should try "levels", "brightness/contrast", and "threshold". Any of those should be able to get stronger blacks and whites for line-art prior to digital painting.

Stray pixels? It sounds like you're trying to erase stuff out or do it all on one layer.

Use layers to overpaint.

Or you could even keep your lines on top. If it's just B/W, set that layer's mode to "Multiply". No need to erase out the white or whatever, because in multiply mode that won't show up. As the old saying goes, "work smarter, not harder".

BTW, double clicking on a layer should unlock it from the background level which is likely if it's a scanned image. That way you can move it above the other layers.

Quote - All in quads?

If possible, yes. Always smooths better that way.

Final product may have a tris version though, client would like to have the model in X-plane in addition to presentation renders. Thing is, I'm not sure if I can figure out Blender well enough for that. (Blender does look like it's improved since I last tried it though, anyone know if there's a Mirai interface mode plugin in addition to default and Maya?) Of course that's another topic in itself.